The flaws of the limit of detection approach

The flaws of the limit of detection approach

Cannabis testing and label failures are huge challenges for the industry, as attorney Ian A. Stewart pointed out in his recent analysis of the “perverse incentives” that cannabis regulations facilitate when products are tested and allowed to pass specifications based on Limit of Detection (LOD).

That is, products that are tested and analyzed for specific contaminants are allowed to “pass” when they are not detected. While this may sound like a reasonable approach to the non-scientist, Stewart clearly understands the flaw in the LOD approach. 

Cannabis regulations exist for many reasons, most of which center on consumer information and consumer safety. Some regulations fall outside normal laboratory processes and leave the laboratory vulnerable to criticism and subject to accusations of providing poor science.

Regulations that not only permit but also demand pass/fail results based on proximity to the LOD create a system that rewards inferior methodology/technology and directs customers away from advanced technology and technical expertise.

These factors compromise the very intention of the regulations.

There is no universally accepted definition for the Limit of Detection. But the general agreement within the scientific community is that the LOD is the lowest concentration of an analyte that can be reliably detected by an instrument.

“Reliably detected” is not equivalent to “reliably quantitated” with bias and precision, a critical point that is often (and unfortunately) overlooked. 

LOD on the road

To provide a realistic analogy, pretend you are driving in an unfamiliar area.

You do not see an exit sign 1 mile away; meanwhile your teenage passenger snidely tells you that he can see the sign. You catch the first glimpse of the sign from a ½ mile away, but you cannot clearly and unambiguously read the words.

Uncertain whether it is the right exit, you decide to continue driving. As you come closer to the sign, it becomes clearer but still seems a little blurry. It’s not until you’re 200 feet away that you clearly read the sign and turn onto the exit ramp. And yes, the teenager is snickering at you!

In this example, your eyes are the instrument, and the exit sign is the target of interest. You do not see the sign until you’re ½ mile away (detection limit) but are still not certain of the target and must travel farther (improve your resolution) to be certain before you commit to an action. Interestingly, the teenager in your car has a better limit of detection at 1 mile than you have at ½ mile! 

Reward sound science

The LOD provides a measurement of the lowest concentration of an analyte that is greater than a sample with no analyte (“blank”). Although not always the case, the LOD often resides below both the lowest point on the calibration curve and the limit of quantitation (LOQ).

Cannabis growers and producers comply with regulations for testing and typically choose laboratory services that will minimize the opportunity for test failures while quickly moving products to the shelves.

Regulations that demand laboratories make decisions in this vague region precariously position laboratories that use state-of-the-art technology with outstanding technical programs to compromise their science to retain customers. Essentially, they could lose business for having equipment that more accurately detects pesticides, heavy metals and other contaminants.

As Stewart explains, state cannabis regulators “should examine and identify reasonable action levels” that will reward sound science and better level the testing marketplace. 

Susan Audino, who holds a doctorate in chemistry, is a chemistry consultant and instructor for the American Association for Laboratory Accreditation. She is based in Delaware.